Presents a method for training and serving LLMs like Qwen3.5 by attaching decision heads to perform classification and judgment tasks
The new form factor supports Node.js 24 on Arm64 and allows standard port-listening servers to run without code rewriting.
The D270 features 32 Xᵉ3P cores and 256 XMX engines, designed for air-cooled, standard PCIe 5.0 data center deployments.
Over 1,300 AI employees signed a letter requesting government support for automated AI development pacing.
Meta introduced the IcePack liquid-cooled rack with 102.4T IceCube switches and open-sourced its MetaRoCE transport protocol.
The update introduced a persistent background daemon that copies the platform package and runs an hourly updater, causing conflicts with package manager installations.
Anthropic's Frontier Red Team estimates that small AI models will reach the capability threshold for autonomous self-replication on commercial IoT devices by late 2026.
The new feature allows Claude to autonomously determine agent topology and orchestration plans, replacing manual coordinator configuration.
GitHub outlines four methods—custom instructions, agent skills, custom agents, and MCP servers—to tailor Copilot's behavior to specific project conventions and workflows.
The method reduces pretraining costs by over 99% and enables controlled comparisons of hybrid architectures like Gated KalmaNet, Gated DeltaNet, and Mamba-2.
Google teams have published 90 peer-reviewed papers with over 90,000 citations and provided over $60 million in funding to health research organizations in 2026.
The new model achieves 13+ tokens/s decoding speed on iPhone 17 Pro Max using Lookahead Expert Scheduling.
The update includes 80 PRs and 12 direct commits from 107 contributors, addressing critical bugs in upgrades, message delivery, and long conversation stability.
The new Muse Wonder series introduces Active Memory and Active Focus features, available now via the Meta Model API and Muse Code.
Integrating the hosted Jev decision model with Qwen 3.8 27B reduced execution API costs by 26.6% while maintaining 100% task accuracy on 360 SOP-Bench cases.
GLM-Legal achieves a 19.2% all-pass rate on Harvey LAB and parity with closed-source models on LegalBench and PLawBench while running on a leaner parameter footprint.
Meta aggregates fragmented idle inference capacity into a shared pool to enhance ad relevance without requiring new hardware.
StudentSim outperforms baselines like GPT-5.4 and Maia2 in chess, writing, and math simulations, achieving 90.5% tutor accuracy in reinforcement learning tests.
The new model achieves 85.9% of the BF16 baseline score while being 13.9× smaller in file size.
The U-Lens method estimates model uncertainty in a single forward pass without resampling or labeled data, outperforming supervised baselines in initial evaluations.
NVIDIA contributed Vera Rubin rack designs to the Open Compute Project, enabling partners to build interoperable AI infrastructure with MGX 1.2.
AI Briefing